> For the complete documentation index, see [llms.txt](https://docs.computesdk.com/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.computesdk.com/platform/benchmarks.md). # Benchmarks Managed benchmarks on the ComputeSDK Platform — org-private and subscribed benchmark runs over the REST API and \`compute bench\` CLI. The platform runs managed benchmarks — multi-phase workloads like **Dax** (clone, install, typecheck) — against registered compute, storage, browser, and AI gateway providers, and publishes the results to each organization's feed. Org benchmarks are private by default; categorized benchmark categories can be subscribed to, which entitles your org to run and read those benchmarks as well. ## What a run looks like A **benchmark** is a workload definition owned by an organization. A **run** is one execution of it against a set of **participants** (the providers or models under test). Each participant's work is carried out by **workers**, which emit telemetry events and artifacts; the platform aggregates those into per-run results and composite scores. ## Reading benchmarks over the API ```bash export COMPUTE_API_KEY="csdk_..." BASE=https://platform.computesdk.com/api/v1 curl $BASE/benchmarks -H "Authorization: Bearer $COMPUTE_API_KEY" curl $BASE/benchmarks/dax -H "Authorization: Bearer $COMPUTE_API_KEY" curl "$BASE/benchmarks/dax/runs?limit=20" -H "Authorization: Bearer $COMPUTE_API_KEY" curl $BASE/benchmarks/dax/runs/ -H "Authorization: Bearer $COMPUTE_API_KEY" curl $BASE/benchmarks/dax/runs//results \ -H "Authorization: Bearer $COMPUTE_API_KEY" ``` `GET /api/v1/benchmarks` pages with `limit`/`offset` (max 500) and returns each benchmark with its owner's org slug — subscribers can tell entitled foreign benchmarks from their own. ### Reading another org's entitled benchmarks Benchmarks you can see but don't own are addressed through the `~` namespace: ```bash curl $BASE/benchmarks/~/computesdk/dax -H "Authorization: Bearer $COMPUTE_API_KEY" curl $BASE/benchmarks/~/computesdk/dax/runs -H "Authorization: Bearer $COMPUTE_API_KEY" curl $BASE/benchmarks/~/computesdk/dax/results -H "Authorization: Bearer $COMPUTE_API_KEY" ``` ### Filtering runs and results `GET /benchmarks/{slug}/runs` and `GET /benchmarks/{slug}/results` take a shared filter: * `participant` — comma-separated slugs; only runs they took part in, and only their rows * `since` / `until` — `YYYY-MM-DD` or an ISO timestamp, on run creation time (a date-only `until` is inclusive of that day) * `timeZone` — IANA zone for date-only bounds, default `UTC` * `steps` — results only; comma-separated step names or `all` for per-step overviews ```bash # Runs e2b took part in since Oct 16 curl "$BASE/benchmarks/dax/runs?participant=e2b&since=2026-10-16" \ -H "Authorization: Bearer $COMPUTE_API_KEY" # One participant's per-step results curl "$BASE/benchmarks/dax/results?participant=e2b&steps=create,install" \ -H "Authorization: Bearer $COMPUTE_API_KEY" ``` Filtered responses echo the applied values back in a `filter` field, so a client can tell the server understood the filter. ### Runs, participants, and workers Within a run: * `runs/{runId}/participants` — the providers/models under test and their per-participant state * `runs/{runId}/participants/{slug}/logs` — participant-level logs * `runs/{runId}/iterations` and `runs/{runId}/progress` — progress and per-iteration data * `runs/{runId}/workers/*` — the worker fleet: events, heartbeats, artifacts, completion. (These are primarily for benchmark executors driving a run, not consumers.) * `runs/{runId}/results`, `results/timeline`, `results/imports`, `results/tasks` — the aggregated output and how it was produced * `runs/{runId}/summary`, `runs/summary`, `runs/day/{day}`, `runs/days` — rollups and day-scoped history `GET /api/v1/feed` returns the org's feed — the same activity stream the dashboard shows. ## `compute bench` Everything above is also reachable through the bench half of the CLI: ```bash compute bench benchmarks # list compute bench runs # list runs compute bench results # query results compute bench iterations # per-iteration data compute bench artifacts # download run artifacts compute bench logs # run logs compute bench export # export results compute bench run # execute a run compute bench check # validate a benchmark definition ``` `runs` and `results` take the same filters as the API: ```bash compute bench results dax --participant e2b --steps all --since 7d compute bench runs list dax --participant e2b --day 2026-10-22 --tz America/Los_Angeles compute bench logs dax --participant e2b ``` See the [CLI reference](/platform/cli.md#compute-bench--the-benchmarks-toolchain) for auth and flags. --- # Agent Instructions This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com. ## Querying This Documentation If you need additional information that is not directly available in this page, you can query the documentation by asking a question. Perform an HTTP GET request on the following URL with the `ask` and `goal` query parameters: ``` GET https://docs.computesdk.com/platform/benchmarks.md?ask=&goal= ``` `ask` is the immediate question: it should be specific, self-contained, and written in natural language. `goal` is what the user is ultimately trying to achieve, the reason they need the answer. Sharing it helps GitBook give you a better, more relevant answer. A goal is most helpful when it describes the outcome the user wants rather than restating the question. For example, with `ask=how do I create an API token`, a goal like `automate deployments from our CI pipeline` lets GitBook tailor the answer to that use case. The response will contain a direct answer to the question and relevant excerpts and sources from the documentation. Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.